1 / 24

How to Fool People to Work on Circuit Lower Bounds

This article explores the idea of fooling people to work on circuit lower bounds by presenting innocent-looking problems seemingly unrelated to proving circuit lower bounds. It discusses the concept of arithmetic circuits, elusive functions, tensor-rank, and lower bounds for arithmetic formulas. The article presents the Tensor-Product Approach as a method for finding super-polynomial lower bounds for arithmetic formulas.

brennen
Download Presentation

How to Fool People to Work on Circuit Lower Bounds

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. How to Fool People to Work on Circuit Lower Bounds Ran Raz Weizmann Institute & Microsoft Research

  2. The Only Barrier for Proving Super-Poly Lower Bounds…

  3. Why Super-Poly Lower Bounds Were Still not Proved ? • Maybe because not enough people are working on it…

  4. The Secret Plan: • Fooling people to work on circuit lower bounds…  • Coming up with innocent looking • clean and simple problems that • are seemingly unrelated to proving • circuit lower bounds, and whose • solution would imply strong circuit • lower bounds

  5. Arithmetic Circuits: • Field: F • Variables: X1,...,Xn • Gates: • Every gate in the circuit computes • a polynomial in F[X1,...,Xn] • Example:(X1¢ X1) ¢ (X2+ 1)

  6. The Holy Grail: • Super-polynomial lower bounds • for circuit or formula size • I will present two innocent looking • problems that imply such bounds

  7. Elusive Functions and Lower Bounds for Arithmetic Circuits

  8. Polynomial Mappings: • f = (f1,...,fm): Cn! Cmis a • polynomial mapping of degreedif • f1,...,fmare polynomials of (total) • degreed • f is explicit if given a monomial M • and index i, the coefficient of M in • fi can be computed in poly time [Val]

  9. The Moments Curve: • f: C ! Cm • f(x) = (x,x2,x3,...,xm) • Fact: 8 affine subspace A ( Cm • 8 :Cm-1! Cm of (total) degree 1,

  10. The Exercise that Was Never Given: • Give an explicit f: C ! Cms.t.: • 8: Cm-1! Cm of degree2, • We require: f of degree · • [R08]:Any explicitf • )super-polynomial lower bounds • for the permanent

  11. Elusive Functions: • f: Cn! Cmis (s,r)-elusive if • 8: Cs! Cm of degreer, • [R08]: explicit constructions of • elusive functions imply lower bounds • for the size of arithmetic circuits

  12. Proof Idea: • Consider : Cs! Cm of degreer, that maps • a circuit to the polynomial computed by it • =polynomials that can be • computed by small circuits. • Proving lower bounds, • Finding points outside • Since • f hits a hard function • Add input variables of f as additional • input variables

  13. Lower Bounds for Depth-d Circuits: • [SS91], [R08]: • Lower bounds of n1+(1/d) • (using elusive functions)

  14. Tensor-Rank and Lower Bounds for Arithmetic Formulas

  15. Tensor-Rank: • A: [n]r! F is of rank 1 if • 9 a1,…,ar : [n] ! F s.t. • A = a1­ a2 ­… ­ ar, that is • A(i1,…,ir) = a1(i1) ¢¢¢ ar(ir) • Rank(A) = Min ks.t. A=A1+…+Ak • where A1,…,Ak are of rank 1 • 8A: [n]r! F Rank(A) · nr-1 • (generalization of matrix rank)

  16. Tensors and Polynomials: • Given A: [n]r! F and n¢r variables • x1,1,…,xr,n define

  17. Tensor-Rank and Arithmetic Circuits: • [Str73]: explicit A:[n]3!F of rank m • )explicit lower bound of (m) • for arithmetic circuits (for fA) • (may give lower bounds of up to(n2)) • (best known bound: (n)) • [R09]: 8 r · logn/loglogn • explicit A:[n]r!F of rank nr(1-o(1)) • )explicit super-poly lower bound • for arithmetic formulas (for fA)

  18. Depth-3 vs. General Formulas: • Tensor-rank corresponds to depth-3 • set-multilinear formulas (for fA) • Corollary: strong enough lower bounds • for depth-3 formulas ) super-poly • lower bounds for general formulas • Folklore:strong enough bounds for depth-4 • circuits)exp bounds for general circuits • [AV08]:any exp bound for depth-4 • circuits)exp bound for general circuits

  19. The Tensor-Product Approach [Str]: • Given A1:[n1]r!F, A2:[n2]r!F • Define A = A1­A2 : [n1¢n2]r ! Fby • A((i1,j1),…,(ir,jr)) = A1(i1,…,ir) ¢ A2(j1,…jr) • For r=2, Rank(A) = Rank(A1)¢Rank(A2) • Is Rank(A) > Rank(A1)¢Rank(A2)/no(1) (8r) ? • YES ) super-poly lower bounds for • arithmetic formulas

  20. The Tensor-Product Approach [Str]: • Given A1:[n1]r!F, A2:[n2]r!F • Define A = A1­A2 : [n1¢n2]r ! Fby • A((i1,j1),…,(ir,jr)) = A1(i1,…,ir) ¢ A2(j1,…jr) • For r=2, Rank(A) = Rank(A1)¢Rank(A2) • Is Rank(A) > Rank(A1)¢Rank(A2)/no(1) (8r) ? • YES ) super-poly lower bounds for • arithmetic formulas • Proof: Let m=n1/r • Take A1,…,Ar:[m]r!Fof high rank • Let A = A1 ­ A2 ­ … ­ Ar : [n]r! F • How do we findA1,…,Arof high rank ? • We fix theirr¢n entries as inputs !

  21. Main Steps of the Proof: • 1) New homogenization and • multilinearization techniques • 2) Defining syntactic-rank of a • formula (bounds the tensor-rank) • 3)8s we find the formula of size s • with the largest syntactic-rank • 4) Compute the largest syntactic-rank of a poly-size formula

  22. Conclusions (of Step 1): • For r · logn/loglogn • 1) super-poly lower bounds for • homogenous formulas ) super-poly • lower bounds for general formulas • 2) super-poly lower bounds for • set-mult formulas ) super-poly • lower bounds for general formulas

  23. Homogenization: • Given a formula C of size s for a • homogenous polynomial f of deg r • give a homogenous formula D for f • [Str73]:D of size sO(log r) • (optimality conjectured in [NW95]) • [R09]:D of size • (where d = product depth of C) • If s=poly(n), and r · logn/loglogn • Size(D)=poly(n)

  24. Thanks!

More Related